One control room for everything your AI does.
SCADA for AI brings supervisory control and data acquisition to AI operations. Watch your agents, workflows, automations and model calls in one place; set the limits they run within; and keep a named person in charge of every consequential decision. It works alongside APIP and the AGS apps, and on your own services too.
Supervision that keeps people in charge
Industrial SCADA lets operators watch and steer a plant from a central room. This does the same job for AI operations: acquire the signals, show the state, hold the limits, raise the alarm and record what people decided.
Acquisition
Collect events from AI agents, co-pilots, automations, model calls and tool use across APIP, the AGS apps and your own services.
Supervision
See every agent's state, volume, confidence, response time, cost and open alerts on one screen.
Control rules
Set limits for confidence, latency, cost, repeated failures, missing policy citations and overdue reviews — and test a rule against real telemetry before it goes live.
Human review
Recommendations queue for a named person to approve, decline or send back. Nothing consequential happens on its own.
Conversational intelligence
Ask a plain question and get an answer drawn from your own figures, with the gaps and the approval point named.
Evidence
Evidence items, recorded decisions and an append-only supervisory log you can export as CSV or JSON.
Have a go without signing in
The figures below are invented and run in your browser. Move the confidence limit to see which agents a control rule would catch — the same test you run against real telemetry inside the app.
AI events (24h)
900
Failed calls
24
Open alerts
7
This rule would raise an alert for 1 of 6 agents: Procurement Risk Screener.
| Agent | State | Risk | Events | Failed | Confidence |
|---|---|---|---|---|---|
| Invoice Triage Agent | live | high | 214 | 9 | 78% |
| Customer Response Co-pilot | live | medium | 186 | 4 | 86% |
| Procurement Risk Screenerwould alert | pilot | high | 96 | 7 | 69% |
| Policy Lens Assistant | live | low | 143 | 1 | 91% |
| Case Routing Automation | live | medium | 178 | 3 | 84% |
| Flow Sentinel Monitor | live | low | 83 | 0 | 88% |
How it is built
- Deployment
- In your AGS workspace at /apip-app/scada, or as a standalone subscription
- Access control
- Organisation membership, role-based permissions, row-level security in the database
- Data acquired
- Event metadata: agent, type, status, confidence, latency, cost, case and policy reference
- Prompt and response text
- Not stored by default — send only the metadata you need to supervise
- Retention
- Held in your organisation's workspace; example data is marked and can be cleared at any time
- Exports
- Telemetry CSV, alert CSV, supervisory log CSV, evidence pack JSON
- AI assistance
- Plain-language summaries and diagnosis, always labelled and never a final decision
- Audit
- Append-only supervisory log of agent, rule, alert, review and export actions
Included, or on its own
With your APIP plan
SCADA for AI is included from the Professional plan and in Enterprise, alongside the rest of the AGS app ecosystem — same workspace, same identity, same audit trail.
On its own
If you only want AI supervision, it can be enabled as a standalone subscription for your organisation. Tell us how many agents and sources you run and we will confirm the fit before you buy anything.
SCADA for AI records what your systems report and what your people decide. It does not certify that a system is compliant, lawful, secure or safe, and it does not replace your legal, audit, security, privacy or safety professionals.
Ready to supervise your AI operations?
Open the control room in your workspace, or talk to us about running it on its own.